Chromatin and Splicing

Abstract

In the past several years, the relationship between chromatin structure and mRNA processing has been the source of significant investigation across diverse disciplines. Central to these efforts was an unanticipated nonrandom distribution of chromatin marks across transcribed regions of protein-coding genes. In addition to the presence of specific histone modifications at the 5′ and 3′ ends of genes, exonic DNA was demonstrated to present a distinct chromatin landscape relative to intronic DNA. As splicing in higher eukaryotes predominantly occurs co-transcriptionally, these studies raised the possibility that chromatin modifications may aid the spliceosome in the detection of exons amidst vast stretches of noncoding intronic sequences. Recent investigations have supported a direct role for chromatin in splicing regulation and have suggested an intriguing role for splicing in the establishment of chromatin modifications. Here we will summarize an accumulating body of data that begins to reveal extensive coupling between chromatin structure and pre-mRNA splicing.

1. Introduction

Unlike the genes of lower eukaryotes, in which protein-coding sequences are typically uninterrupted, genes of higher metazoans are characterized by a large number of coding exons separated by long stretches of noncoding introns. As genes are transcribed into mRNA, introns are excised by the megadalton spliceosome complex, which recognizes short consensus sequences at intron–exon boundaries [1]. While introns were initially dubbed as “junk DNA,” it is increasingly evident that exon–intron architecture serves as a critical platform for transcriptome diversification via alternative pre-mRNA splicing. Cassette exons thus represent an important aspect of proteome complexity in higher organisms, and current estimates indicate that greater than 90 % of human genes engage in alternative splicing [2, 3]. However, the evolutionary drive for transcriptome expansion has posed the spliceosome with an increasingly difficult task as intron lengths have increased and splice site strengths have weakened [4]. Alternative pre-mRNA splicing adds an additional layer of complexity in that splice site recognition must be diversified in a context-dependent manner. To accomplish regulated transcript production within a multivariable framework, pre-mRNA splicing is coordinated at multiple levels. In addition to regulation via RNA-binding protein recognition of cis-elements encoded within pre-mRNA [5], the rate of RNA polymerase II (pol II) transcription elongation and chromatin structure contribute to splice site recognition [6]. Rather than operating independently, these processes are highly integrated as a result of co-transcriptional pre-mRNA splicing [7, 8]. The basic mechanisms of alternative splicing regulation via RNA-binding proteins and evidence for co-transcriptional splicing are discussed elsewhere in this volume. Here, we will focus directly on the accumulating evidence in support of a role for chromatin structure in splicing regulation.

2. Chromatin and Co-transcriptional Splicing

A central tenet to the relationship between chromatin structure and alternative splicing is that the majority of splicing in higher eukaryotes occurs co-transcriptionally, while the nascent message is still tethered to the template DNA. This allows for several layers of coupling between the transcription and splicing machineries. Initial efforts to address coupling focused on the potential association between RNA-binding proteins and the pol II carboxy-terminal domain (CTD), such that the factors could be efficiently transferred to the nascent transcript during the process of transcription [9]. Combinatorial association of these factors could in principle influence splicing decisions [5]. Co-transcriptionality further allows for kinetic regulation of splicing decisions. In work pioneered in the Kornblihtt group, it was shown that the rate of transcription elongation impacts splicing decisions such that weak exons are more likely to be excluded from spliced mRNA in response to a rapid elongation rate [10]. While the kinetic model has now been validated in a variety of systems, the physiological barriers to pol II elongation remained comparatively elusive until recently. Genome-wide profiling of chromatin modifications revealed the transcribed DNA template itself as a potential modulator of elongation rate or other aspects of splicing regulation. Intragenic DNA presents a distinct chromatin landscape relative to intergenic DNA, and more importantly to this discussion, exonic DNA presents unique features relative to intronic DNA [11]. These observations raised the possibility that the chromatin structure of transcribed genes may aid the spliceosome in the process of exon definition. In this section, we expand on these themes and examine the various evidences and mechanisms for chromatin-directed splicing.

2.1. Evidence for Co-transcriptional Splicing

While pre-mRNA splicing was initially envisioned as a distinct cellular process that occurred subsequent to the completion of transcription, evidence for co-transcriptional splicing quickly mounted. In a widely cited landmark study, electron micrographs of chromosomal spreads from Drosophila embryos provided a visual demonstration of spliceosome assembly on nascent mRNA, while the RNA was still tethered to the template DNA [12, 13]. Further, indirect evidence of functional cross talk between transcription and pre-mRNA splicing came from studies wherein protein-coding genes were placed downstream of RNA polymerase I or RNA polymerase III promoters [14–17]. These studies indicated that pol I and pol III promoters led to efficient mRNA transcription, but the resulting RNAs were poorly spliced, highlighting an obligatory functional connection between pol II and the splicing machinery [14–17]. Similar results were obtained following deletion of the pol II CTD [18]. The CTD consists of multiple repeats (52 in human) of evolutionary conserved heptapeptide (YSPTSPS) sequences that are dynamically phosphorylated at the different stages of transcription [19]. Transient transfection of CTD-deleted pol II completely abrogated splicing of a β-globin reporter [18], suggesting an important role for the CTD in co-transcriptional splicing. This notion was further strengthened with the development of fluorescent microscopy. Misteli et al. used live cell imaging to demonstrate the relocalization of a fluorescently labeled splicing factor from nuclear speckles to sites of new transcription initiated from a β-tropomyosin minigene [20]. Furthermore, they and others have showed that such splicing factor mobilization is dependent on transcription through RNA polymerase II with an intact CTD [21–24]. More recently, several kinetic studies have shown that the vast majority of splicing in higher eukaryotes occurs co-transcriptionally, in a general 5′–3′ order. Altogether, these studies provide a rationale framework for coupling between elements at the transcribed DNA template and the splicing machinery.

2.2. Kinetic Regulation of Splicing

Almost immediately preceding the formal demonstration of co-transcriptional splicing, Kornblihtt and others revealed an intriguing new connection between these two processes that would ultimately pave the way for a new era in the alternative splicing field. In what would come to be known as the “kinetic” hypothesis, the rate of pol II elongation was implicated in alternative splicing decisions. As a first hint of things to come, it was demonstrated that promoter identity influences splicing decisions. Swapping promoters in minigene constructs resulted in altered splicing of weak exons [25], and recruitment of the splicing factor SF2/ASF (now SRSF1) to enhancers was shown to be promoter dependent [26]. These promoter-related splicing effects were ultimately attributed to elongation rate [27]. In support of a kinetic connection, the Smith group further found that inserting a binding site for the zinc-finger protein MAZ downstream of a weak exon led to pol II pausing and increased inclusion of the exon in spliced mRNA [28]. It was thus proposed that alternative splicing decisions are influenced by a temporal window of opportunity. Subsequent work fully established this notion and ultimately revealed a connection to chromatin (discussed below). Focusing on the fibronectin gene, inhibiting pol II elongation rate through use of a chemical inhibitor, 5,6-dichloro-1-beta-D ribofuranosylbenzimidazole (DRB), increased inclusion of the weak EDI exon, whereas increasing elongation rate through treatment with the histone deacetylase inhibitor trichostatin A (TSA) decreased exon inclusion [71]. A similar increase in EDI inclusion was seen in response to transcription with α-amanitin-resistant pol II with reduced elongation rate [10]. The global relevance of kinetic coupling between transcription and pre-mRNA splicing was later solidified through several genome-wide studies. Consistent throughout these studies, a slow elongation rate was associated with weak exon inclusion. For example, inhibition of pol II elongation with DRB or camptothecin in activated human T cells favored inclusion of weak exons in a subset of genes that were enriched in RNA processing and apoptosis pathways [29]. Similarly, hyperphosphorylation of the pol II CTD by ultraviolet irradiation (UV) and consequent inhibition of elongation rate induced alternative splicing of many genes involved in the DNA damage response [30].

Various models have been put forth to address the physiological signals for variable pol II elongation. For example, polymerase itself may be altered due to differential CTD phosphorylation or interaction with accessory factors [31] (Chapter 6). Alternatively, recent evidence suggests that the chromatin template of transcribed genes may provide polymerase with signals that locally regulate elongation rate, or provide direct information to the spliceosome regarding the location of exons. We expand on both these modes of chromatin-regulated splicing below.

2.3. Chromatin-Mediated Regulation of Splicing

Building on the accumulating evidence in support of co-transcriptional mRNA processing [32, 33], the first hints of a role for chromatin in splicing decisions date back to the early 1990s. Beckmann and Trifonov unexpectedly discovered that the average distance between the 3′ and 5′ splice sites flanking an exon followed a periodic pattern that was very close to the length covered by a single nucleosome [34]. Based on these results, they rationalized that placement of nucleosomes according to exon–intron boundaries may reflect an unanticipated role for chromatin structure in pre-mRNA splicing [34]. At the same time, an involvement for chromatin was also suspected when integrated copies of the adenovirus genome at distinct genomic locations in the same nuclei yielded different splicing outcomes [35]. Given that the viral genomes and cellular contexts were identical, the authors speculated that the chromatin structure at the integration site was responsible for the distinct splicing patterns, possibly through influencing the pol II elongation rate [35]. While these ideas gained momentum in the following years, it wasn’t until the advent of genome-wide sequencing data showing distinct chromatin patterns at exonic relative to intronic sequence that chromatin was solidified as a genuine contributor to splicing regulation. Genome-wide studies revealed that exons show a higher rate of nucleosome occupancy, specific histone modifications, and elevated DNA methylation relative to introns, raising the possibility that chromatin may aid the spliceosome in the process of exon definition. We discuss each of these associations in turn below.

2.3.1. Nucleosomes

A fundamental aspect of gene regulation in eukaryotes is the packaging of DNA into higher-order structures. In addition to maintaining genome integrity, this allows for the control of gene expression through the adoption of either transcriptionally inaccessible heterochromatin or relatively open euchromatin [36]. The basic building block of chromatin is the nucleosome, which is composed of approximately 146 base pairs of DNA wound around an octamer of histone proteins [37]. Canonical nucleosomes include two copies each of H2A, H2B, H3, and H4 histones [37]. Nucleosomes are inherent barriers to pol II elongation, as has been effectively demonstrated in vitro [38–40]. In order to accomplish efficient elongation in vivo, a variety of proteins cooperate to disassemble nucleosomes in front of elongating pol II and reestablish them in its wake [11, 41, 42]. Nucleosome turnover is a critical aspect of transcription fidelity, as nucleosome depletion facilitates unwanted cryptic transcription [43–47]. Given these very basic roles for nucleosomes in transcription, it was surprising to find a nonrandom intragenic positioning pattern across the genome: nucleosome occupancy is elevated at exons relative to introns, irrespective of gene expression status [48–50]. Considering the in vitro data demonstrating that nucleosomes are barriers to pol II elongation, these observations suggested that nucleosomes promote exon definition by facilitating transient pol II pausing and consequent spliceosome assembly. While not formally demonstrated due to difficulties in depleting nucleosomes without untoward effects on gene expression, several studies support this premise. For example, exons with weak splice sites show higher nucleosome density compared to constitutive exons, and pseudoexons with strong splice sites are nucleosome depleted [48]. Furthermore, the average size of a mammalian exon (145 base pairs) is similar to the length of DNA wrapped within a single nucleosome [48, 51]. Additionally, a recent study suggested a role for variant histone incorporation in splicing regulation. Depletion of the mammalian-specific H2A variant, H2A.Bbd, which is preferentially found in transcribed sequences, led to a global decrease in splicing efficiency [52]. As several proteins involved in RNA processing as well as major splicing components were coprecipitated with H2A.Bbd [52–55], these studies hint at a direct role for nucleosomes in splicing regulation. This notion is further reinforced by evidence implicating posttranslational modification of histone tails in splicing regulation, as described below.

An additional twist in the interplay between nucleosomes and splicing is related to nucleotide content. Exonic sequences tend to be GC-rich as compared to introns, which inherently favors nucleosome deposition at exons [56, 57]. Nucleotide bias has been implicated in alternative splicing regulation [58], and a recent study identified a role for the DBIRD complex (composed of deleted in breast cancer 1 [DBC1] and ZNF236) in the exclusion of AT-rich exons. Through a yet undefined mechanism, DBIRD interacts with pol II and facilitates efficient passage through AT-rich sequence. In the absence of DBIRD, polymerase stalls at the weak exons and leads to increased inclusion [59]. This study highlights an additional emerging theme: splicing, chromatin, and transcription are highly intertwined.

2.3.2. Modification of Histone Tails

The advent of high-throughput deep sequencing following chromatin immunoprecipitation (ChIP-seq) has allowed for the dissection of the intragenic epigenome across diverse species, including human, mice, C. elegans, and Drosophila [50, 51, 60, 61]. Several histone marks were found to be particularly prevalent on exonic sequences, including trimethylation of H3 lysine 36 (H3K36me3) [47, 62–65], dimethylation of H3 lysine 27 (H3K27me2) [49, 64], and monomethylation of H3 lysine 79 (H3K79me1) and H2B lysine 5 (H2BK5me1) [62, 63]. In contrast, H3K9 methylation is depleted at exons [66]. Intriguingly, exonic enrichment is not evenly distributed across gene bodies. For example, H3K36me3 levels rise into gene bodies, whereas H3K4me3 is found near transcription start sites [60]. H3K36me3, in particular, has received significant attention as it is exclusively found at actively transcribed genes, suggesting an important role in pre-mRNA processing [60]. Acknowledging that histone modification measurements must be adjusted for the overall increase in nucleosome content at exonic sequences, H3K36me3 enrichment at exons persists after nucleosome correction [64]. While the genome-wide associations have largely focused on histone methylation, additional modifications to histone tails have the potential to influence splicing. For example, a recent study in yeast reported elevated monoubiquitylation of H2B lysine 123 (H2BK123ub1) in introns of transcribed genes, and disruption of H2BK123ub1 altered the distribution of H3K6me3 [67], suggesting functional antagonism in exon–intron definition through specific posttranslational modifications of histone tails.

A direct role for histone modifications in exon definition is supported by a variety of studies involving modulation of specific histone posttranslational marks. As a result, two non-exclusive potential mechanisms by which chromatin can influence alternative splicing decisions have been proposed: (1) local alteration of pol II elongation rate and (2) site-specific recruitment of RNA-binding proteins through interaction with chromatin-binding proteins. Examples of both modes of regulation exist and are largely intertwined as described below.

Histone Modifications in Kinetic Regulation

The strongest evidence in favor of a role for chromatin in kinetic regulation of splicing comes from studies in which exogenous stimuli led to global or local changes in chromatin structure and associated changes in splicing. In general, acetylation of histone tails is associated with an open chromatin context and efficient pol II processivity [68]. Several studies from the Kornblihtt group have shown that tipping the “accessibility” balance in either direction alters exon inclusion of both model genes and genome wide. For example, membrane depolarization of neuronal cells led to increased H3K9 acetylation (H3K9ac) specifically within the vicinity of exon 18 of the NCAM gene and promoted exclusion of the exon from spliced mRNA [69]. This exon appears to be particularly susceptible to kinetic regulation, as evidenced through use of a mutant polymerase with reduced elongation rate. In addition, globally increasing acetylation with the histone deacetylase inhibitor trichostatin A (TSA) decreased exon 18 inclusion [70]. A similar decrease in the fibronectin EDI exon was seen following TSA treatment [71]. In contrast, the opposing, repressive modification, H3K9 methylation, has been shown to mediate exon inclusion in a number of systems. The Muchardt group defined a role for H3K9me3 in alternative splicing of CD44 pre-mRNA and was further able to uncover a potential physiological link to polymerase pausing. In response to phorbol 12-myristate 13-acetate (PMA) treatment, CD44 transcripts show increased inclusion of nine tandem alternative exons, which is associated with increased pol II occupancy and a local increase in H3K9me3 detection [72, 73]. Stimulation also resulted in increased detection of the H3K9me3-interacting chromodomain protein HP1γ at the alternative exons. Remarkably, RNAi-mediated depletion of HP1γ abrogated both pol II accumulation and variant exon inclusion [74]. This study suggested that H3K9me3-associated HP1γ somehow bridged the processes of transcription and splicing. Indeed, it has been reported that HP1γ is enriched at hundreds of active genes and promotes co-transcriptional splicing through recruitment of the spliceosomal protein U1–70K and SR family protein SRSF1 [75, 76].

Additional evidence supporting a role for H3K9 methylation in the kinetic regulation of splicing comes from studies involving exogenous introduction of siRNAs directed against intragenic sequence. Analogous to Argonaute protein-dependent transcriptional gene silencing (TGS), wherein siRNA directed against promoter DNA triggers gene silencing through local heterochromatin formation [77–80], extension of TGS into intragenic sequence mediates chromatin changes that locally modulate pol II elongation without affecting overall gene expression. For example, exogenous siRNAs targeted against an intronic sequence proximal to the fibronectin EDI exon in human cells promoted an AGO1-dependent local increase in the repressive chromatin marks H3K9me2 and H3K27me3 and increased exon inclusion [81]. The authors further demonstrated that the altered chromatin structure resulted in HP1α recruitment to the region, suggesting a similar bridging effect as described for HP1γ above. A recent study in the CD44 model system also directly implicated the Argonaute proteins AGO1 and AGO2 in bridging interactions. AGO1 and AGO2 interact with splicing factors and are recruited to the CD44 variant exons in a Dicer and HP1γ dependent–manner, culminating in increased exon inclusion through reduced pol II processivity [82]. These complex associations were reverberated in several studies from the Kennedy laboratory examining the Argonaute-related nuclear RNAi defective (NRDE) protein-dependent intragenic TGS pathway in C. elegans. Both exogenous and endogenous siRNA-associated recruitment of NRDE factors were found to promote accumulation of H3K9me3 and inhibited pol II elongation in C. elegans [83, 84]. Furthermore, endogenous siRNAs directed NRDE-1 to interact with both chromatin and pre-mRNA, thereby revealing a conserved role for Argonaute proteins in connecting these nuclear processes [84, 85].

As is evident throughout these studies, post-transcriptional modification of H3K9 seems to have a central role in the kinetic regulation of splicing. It is worth noting that these observations are somewhat at odds with the intergenic role of H3K9. Trimethylation of H3K9 is a classic feature of heterochromatin formation and is associated with repeat elements and otherwise silenced areas of the genome [86]. However, intragenic H3K9me3 is not strictly associated with transcriptional repression [87]. These studies suggest that H3K9 may mediate context-dependent effects on transcription and/or splicing. It will certainly be interesting to examine the role of this modification in splicing in greater detail in the coming years.

Histone Modifications in Adaptor Function

While several aspects of bridging from chromatin to RNA were highlighted in the discussion of kinetic regulation above, it is unlikely that all alternative exons are strictly under kinetic regulation. Indeed, studies focused on the role of H3K36 and H3K4 methylation in splicing reveal a more general role for chromatin modifications as adaptors for RNA-binding protein recruitment. Interestingly, as shown for H3K36me3, the same chromatin modification can be associated with distinct splicing outcomes dependent on the interacting factors. For example, the Misteli laboratory has shown that H3K36me3 is associated with exclusion of a subset of PTB-dependent exons. PTB is recruited to these exons through interaction with the H3K36me3-binding protein, MRG15 [88]. As a proof of principle, depletion of H3K36me3 levels through RNAi against the K36 methyltransferase, Setd2, increased the inclusion level of these exons [88]. In contrast, the Bickmore laboratory has shown that H3K36me3 is associated with inclusion of a subset of exons due to Psip1-/Ledgf-dependent recruitment of the splicing factor SRSF1. In a strikingly similar mechanism, Psip1 interacts with H3K36me3 and recruits the positive acting splicing factor to a subset of exons, and SRSF1 localization and splicing are altered in Psip1 null cells [89]. These contrasting studies illustrate the clear involvement of additional context-dependent factors that remain to be identified.

Adaptor function has also been demonstrated for H3K4me3, which is enriched at the 5′ ends of active genes. Biochemical purification identified CHD1 as an H3K4me3-interacting protein and CHD1 was also found to interact with the spliceosomal proteins U2 snRNP [90]. Depletion of either CHD1 or H3K4me3 through RNAi reduced U2 association with chromatin and reduced pre-mRNA splicing efficiency [90]. Similarly, the histone 3 acetyltransferase, GCN5, promotes co-transcriptional U2 snRNP recruitment [91], hinting at a possible adaptor function. Altogether, the sum of these studies demonstrates a clear role for chromatin structure in constitutive and alternative splicing regulation, both through kinetic and adaptor mechanisms.

2.4. DNA Methylation

As introduced above, DNA methylation also shows a nonrandom intragenic distribution pattern: methylation levels are significantly enhanced at exonic relative to intronic sequences. However, unlike promoter DNA methylation, which is associated with gene silencing [92], methylation within gene bodies does not have a clear relationship to gene expression levels [93]. Instead, genome-wide methylome analyses in lower eukaryotes foreshadowed a potential role for DNA methylation in splicing regulation. Comparative methylome analyses from the Jacobsen and Zilberman laboratories showed that the acquisition of DNA methylation predates the divergence of plant and animal lineages, and revealed conserved enrichment at exonic sequences relative to introns [94, 95]. In comparing genetically identical honeybee castes, the Maleszka laboratory further showed that differences in queen versus worker bee methylome patterns correlate with changes in alternative splicing patterns. Strikingly, they also found that the low level of DNA methylation is seemingly restricted to exons and is depleted at intronless genes [96]. Subsequent work in additional insect model systems has confirmed an association between intragenic methylation, alternative splicing, and phenotypic diversity [97–99]. While mammalian methylomes are comparatively complex and widespread intergenic methylation is found, these intragenic features are highly conserved [100, 101]. High-resolution bisulfite sequencing of the human genome validated the enrichment of exonic methylation and revealed sharp transitions at exon–intron junctions [102]. A reanalysis of several genome-wide human methylome and RNA-seq datasets established that methylation levels correlate with alternative splicing in human cells. Included exons showed an overall higher level of DNA methylation relative to excluded exons, suggesting a direct role for DNA methylation in exon definition. These associations persisted even after correcting for increased nucleosome and GC content at exons relative to introns [93].

The conserved association between exonic DNA methylation and alternative splicing across diverse taxa suggests regulated mechanisms for the establishment and removal of methylation patterns. While the mechanisms by which DNA methyltransferases (DNMTs) are targeted to exons at distinct stages in development remain unknown, recent studies have begun to reveal a basis for variable DNA methylation. 5-methylcytosine (5-mC) can be converted to 5-hydroxymethylcytosine (5-hmC) through the activity of the TET family of proteins. 5-hmC can stably persist in the genome, or can be further converted to unmethylated cytosine through additional oxidation. Notably, bisulfite sequencing is unable to distinguish 5-mC from 5-hmC, and recent studies aimed at specifically deciphering the 5-hmC methylome have found an overlapping distribution pattern: 5-hmC is also enriched at exons relative to introns [103, 104]. Furthermore, 5-hmC levels were also shown to undergo sharp transitions at exon–intron boundaries in the brain, and alternative exons showed an overall lower level of 5-hmC relative to constitutive exons. Curiously, non-neural tissues showed more 5-mC at exon–intron boundaries [105]. These studies suggest that tissue-specific changes in the ratio of 5-mC to 5-hmC may represent a novel mode of alternative splicing regulation.

While the accumulation of these genome-wide data over the last several years strongly suggested a fundamental role for DNA methylation in exon definition, potential mechanisms had remained more elusive. Given that the majority of splicing occurs co-transcriptionally, possibilities included direct impact of DNA methylation on splicing through kinetic regulation or indirect regulation through variable interaction with auxiliary factors. We recently provided evidence for the latter possibility. Through our analysis of alternative splicing of CD45 pre-mRNA, we determined that inclusion of variable exon 5 is mediated by reciprocal binding of the zinc-finger protein, CTCF, and 5-methylcytosine. The binding of CTCF to CD45 DNA acts as a transient barrier to pol II elongation, which kinetically favors spliceosome assembly at the weak splice sites. In contrast, DNA methylation acts to evict CTCF and thereby abolishes pol II pausing and exon 5 inclusion. CTCF-ChIP-seq and RNA-seq in CTCF-depleted cells verified CTCF to be a global regulator of alternative splicing [106]. This study provided the first mechanistic link between DNA methylation and alternative splicing. A similar effect was recently shown for the zinc-finger protein VEZF1: binding of VEZF1 to DNA promotes pol II pausing and results in alternative splicing of a subset of genes [107]. Interestingly, like CTCF, VEZF1 interaction with DNA protects against DNA methylation [108]. In addition, VEZF1 interacts with MRG15, which was previously implicated to be a chromatin-binding adaptor between H3K36me3 and PTB as described above [88, 107]. Together, these studies suggest a basic role for intragenic binding of zinc-finger proteins in kinetically regulated splicing. We predict that many additional examples of DNA-binding regulators of splicing will be revealed in the coming years.

It should further be noted that we found CD45 exon 5 methylation to be developmentally regulated. Naïve peripheral lymphocytes show enhanced CTCF binding, whereas mature lymphocytes show enhanced exon 5 methylation. Thus, mechanisms certainly exist that promote exon 5 methylation in a stage-specific manner. Support for active remodeling of intragenic methylation can also be found in the honeybee genome. As mentioned above, genetically identical queen and worker honeybee show distinct methylation profiles. Remarkably, RNAi-mediated depletion of the de novo methyltransferase DNMT3 generated adult bees with queen-like characteristics, and queen-specific isoforms have been defined that are distinguished by unique overlapping exonic DNA methylation [96, 109]. Given that DNA methylation is not strictly associated with exon inclusion or exclusion [110], it is possible that intragenic DNA methylation plays a fundamental role in developmentally regulated alternative splicing through association with a complex network of methyl-resistant and methyl-sensitive DNA-binding proteins.

3. Splicing Reciprocally Modulates Chromatin Structure

While the role of chromatin structure in splicing is now well established, an emerging area of coupling is recent evidence showing that splicing can reciprocally influence chromatin modifications. Independent publications from the Carmo-Fonseca and Bentley laboratories indicated that H3K36me3 deposition over gene bodies is splicing dependent [111, 112]. The Bentley laboratory showed that mutation of 3′ splice sites in the upstream introns of a β-globin reporter resulted in repositioning of H3K36me3 from the 5′ to the 3′ end of the reporter. Globally inhibiting splicing with spliceostatin A also resulted in H3K36me3 redistribution further 3′ into genes [111]. The Carmo-Fonseca group further showed that splicing promotes recruitment of the H3K36 methyltransferase HYPB/Setd2 to gene bodies. Inhibition of splicing globally reduced H3K36me3, whereas activating splicing of a model gene had the opposite effect. Furthermore, intronless genes show lower H3K6me3 levels irrespective of expression status [112], implicating splicing rather than general transcription in Setd2 recruitment. The notion that pre-mRNA processing can influence histone modification is further strengthened by the demonstration that the RNA-binding Hu proteins can modulate histone acetylation. The binding of Hu proteins to target sites on mRNA flanking alternative exons led to local inhibition of histone deacetylase 2 (HDAC2) activity and increased histone acetylation and exon exclusion [113]. Altogether, these studies suggest that spliceosomal components actively connect histone-modifying enzymes to transcription through yet unknown mechanisms.

4. Concluding Remarks

As highlighted throughout this chapter, the last several years have revealed an extensive network of coupling between pre-mRNA splicing and chromatin. Studies of model genes and genome-wide analyses have revealed exonic epigenetic signatures. Some of these signatures, such as nucleosomes [48–50], are found independent of transcription status, whereas others, such as H3K36me3, are transcription and splicing dependent [111, 112]. Chromatin modifications have been shown to influence spliceosome assembly through kinetic regulation of pol II elongation and through recruitment of splicing factors to their required sites of action [11]. An area that will be particularly interesting to follow in the coming years is how these changes in chromatin structure are modulated during development. While the current studies have focused on the chromatin template, pol II, and the spliceosome, in reality, a vast network of remodelers is required to effect chromatin changes. For example, how are constitutively expressed histone- and DNA-modifying enzymes targeted to intragenic sequences at specific stages in development? In addition, do additional chromatin-associated factors that are critical for transcription, such as histone chaperones, have a role in splicing regulation? Certainly, our current understanding of chromatin-mediated mRNA splicing field is hazy at best, and future research in this area is likely to be full of surprises.